readline/support/wcwidth.c

   1 /*
   2  * This is an implementation of wcwidth() and wcswidth() (defined in
   3  * IEEE Std 1002.1-2001) for Unicode.
   4  *
   5  * http://www.opengroup.org/onlinepubs/007904975/functions/wcwidth.html
   6  * http://www.opengroup.org/onlinepubs/007904975/functions/wcswidth.html
   7  *
   8  * In fixed-width output devices, Latin characters all occupy a single
   9  * "cell" position of equal width, whereas ideographic CJK characters
  10  * occupy two such cells. Interoperability between terminal-line
  11  * applications and (teletype-style) character terminals using the
  12  * UTF-8 encoding requires agreement on which character should advance
  13  * the cursor by how many cell positions. No established formal
  14  * standards exist at present on which Unicode character shall occupy
  15  * how many cell positions on character terminals. These routines are
  16  * a first attempt of defining such behavior based on simple rules
  17  * applied to data provided by the Unicode Consortium.
  18  *
  19  * For some graphical characters, the Unicode standard explicitly
  20  * defines a character-cell width via the definition of the East Asian
  21  * FullWidth (F), Wide (W), Half-width (H), and Narrow (Na) classes.
  22  * In all these cases, there is no ambiguity about which width a
  23  * terminal shall use. For characters in the East Asian Ambiguous (A)
  24  * class, the width choice depends purely on a preference of backward
  25  * compatibility with either historic CJK or Western practice.
  26  * Choosing single-width for these characters is easy to justify as
  27  * the appropriate long-term solution, as the CJK practice of
  28  * displaying these characters as double-width comes from historic
  29  * implementation simplicity (8-bit encoded characters were displayed
  30  * single-width and 16-bit ones double-width, even for Greek,
  31  * Cyrillic, etc.) and not any typographic considerations.
  32  *
  33  * Much less clear is the choice of width for the Not East Asian
  34  * (Neutral) class. Existing practice does not dictate a width for any
  35  * of these characters. It would nevertheless make sense
  36  * typographically to allocate two character cells to characters such
  37  * as for instance EM SPACE or VOLUME INTEGRAL, which cannot be
  38  * represented adequately with a single-width glyph. The following
  39  * routines at present merely assign a single-cell width to all
  40  * neutral characters, in the interest of simplicity. This is not
  41  * entirely satisfactory and should be reconsidered before
  42  * establishing a formal standard in this area. At the moment, the
  43  * decision which Not East Asian (Neutral) characters should be
  44  * represented by double-width glyphs cannot yet be answered by
  45  * applying a simple rule from the Unicode database content. Setting
  46  * up a proper standard for the behavior of UTF-8 character terminals
  47  * will require a careful analysis not only of each Unicode character,
  48  * but also of each presentation form, something the author of these
  49  * routines has avoided to do so far.
  50  *
  51  * http://www.unicode.org/unicode/reports/tr11/
  52  *
  53  * Markus Kuhn -- 2007-05-26 (Unicode 5.0)
  54  *
  55  * Permission to use, copy, modify, and distribute this software
  56  * for any purpose and without fee is hereby granted. The author
  57  * disclaims all warranties with regard to this software.
  58  *
  59  * Latest version: http://www.cl.cam.ac.uk/~mgk25/ucs/wcwidth.c
  60  */
  61
  62 #ifdef __GO32__
  63 #  include <wctype.h>
  64 #endif
  65
  66 #include <wchar.h>
  67
  68 struct interval {
  69   int first;
  70   int last;
  71 };
  72
  73 /* auxiliary function for binary search in interval table */
  74 static int bisearch(wchar_t ucs, const struct interval *table, int max) {
  75   int min = 0;
  76   int mid;
  77
  78   if (ucs < table[0].first || ucs > table[max].last)
  79     return 0;
  80   while (max >= min) {
  81     mid = (min + max) / 2;
  82     if (ucs > table[mid].last)
  83       min = mid + 1;
  84     else if (ucs < table[mid].first)
  85       max = mid - 1;
  86     else
  87       return 1;
  88   }
  89
  90   return 0;
  91 }
  92
  93
  94 /* The following two functions define the column width of an ISO 10646
  95  * character as follows:
  96  *
  97  *    - The null character (U+0000) has a column width of 0.
  98  *
  99  *    - Other C0/C1 control characters and DEL will lead to a return
 100  *      value of -1.
 101  *
 102  *    - Non-spacing and enclosing combining characters (general
 103  *      category code Mn or Me in the Unicode database) have a
 104  *      column width of 0.
 105  *
 106  *    - SOFT HYPHEN (U+00AD) has a column width of 1.
 107  *
 108  *    - Other format characters (general category code Cf in the Unicode
 109  *      database) and ZERO WIDTH SPACE (U+200B) have a column width of 0.
 110  *
 111  *    - Hangul Jamo medial vowels and final consonants (U+1160-U+11FF)
 112  *      have a column width of 0.
 113  *
 114  *    - Spacing characters in the East Asian Wide (W) or East Asian
 115  *      Full-width (F) category as defined in Unicode Technical
 116  *      Report #11 have a column width of 2.
 117  *
 118  *    - All remaining characters (including all printable
 119  *      ISO 8859-1 and WGL4 characters, Unicode control characters,
 120  *      etc.) have a column width of 1.
 121  *
 122  * This implementation assumes that wchar_t characters are encoded
 123  * in ISO 10646.
 124  */
 125
 126 int mk_wcwidth(wchar_t ucs)
 127 {
 128   /* sorted list of non-overlapping intervals of non-spacing characters */
 129   /* generated by "uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c" */
 130   static const struct interval combining[] = {
 131     { 0x0300, 0x036F }, { 0x0483, 0x0486 }, { 0x0488, 0x0489 },
 132     { 0x0591, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 },
 133     { 0x05C4, 0x05C5 }, { 0x05C7, 0x05C7 }, { 0x0600, 0x0603 },
 134     { 0x0610, 0x0615 }, { 0x064B, 0x065E }, { 0x0670, 0x0670 },
 135     { 0x06D6, 0x06E4 }, { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED },
 136     { 0x070F, 0x070F }, { 0x0711, 0x0711 }, { 0x0730, 0x074A },
 137     { 0x07A6, 0x07B0 }, { 0x07EB, 0x07F3 }, { 0x0901, 0x0902 },
 138     { 0x093C, 0x093C }, { 0x0941, 0x0948 }, { 0x094D, 0x094D },
 139     { 0x0951, 0x0954 }, { 0x0962, 0x0963 }, { 0x0981, 0x0981 },
 140     { 0x09BC, 0x09BC }, { 0x09C1, 0x09C4 }, { 0x09CD, 0x09CD },
 141     { 0x09E2, 0x09E3 }, { 0x0A01, 0x0A02 }, { 0x0A3C, 0x0A3C },
 142     { 0x0A41, 0x0A42 }, { 0x0A47, 0x0A48 }, { 0x0A4B, 0x0A4D },
 143     { 0x0A70, 0x0A71 }, { 0x0A81, 0x0A82 }, { 0x0ABC, 0x0ABC },
 144     { 0x0AC1, 0x0AC5 }, { 0x0AC7, 0x0AC8 }, { 0x0ACD, 0x0ACD },
 145     { 0x0AE2, 0x0AE3 }, { 0x0B01, 0x0B01 }, { 0x0B3C, 0x0B3C },
 146     { 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B43 }, { 0x0B4D, 0x0B4D },
 147     { 0x0B56, 0x0B56 }, { 0x0B82, 0x0B82 }, { 0x0BC0, 0x0BC0 },
 148     { 0x0BCD, 0x0BCD }, { 0x0C3E, 0x0C40 }, { 0x0C46, 0x0C48 },
 149     { 0x0C4A, 0x0C4D }, { 0x0C55, 0x0C56 }, { 0x0CBC, 0x0CBC },
 150     { 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 }, { 0x0CCC, 0x0CCD },
 151     { 0x0CE2, 0x0CE3 }, { 0x0D41, 0x0D43 }, { 0x0D4D, 0x0D4D },
 152     { 0x0DCA, 0x0DCA }, { 0x0DD2, 0x0DD4 }, { 0x0DD6, 0x0DD6 },
 153     { 0x0E31, 0x0E31 }, { 0x0E34, 0x0E3A }, { 0x0E47, 0x0E4E },
 154     { 0x0EB1, 0x0EB1 }, { 0x0EB4, 0x0EB9 }, { 0x0EBB, 0x0EBC },
 155     { 0x0EC8, 0x0ECD }, { 0x0F18, 0x0F19 }, { 0x0F35, 0x0F35 },
 156     { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 }, { 0x0F71, 0x0F7E },
 157     { 0x0F80, 0x0F84 }, { 0x0F86, 0x0F87 }, { 0x0F90, 0x0F97 },
 158     { 0x0F99, 0x0FBC }, { 0x0FC6, 0x0FC6 }, { 0x102D, 0x1030 },
 159     { 0x1032, 0x1032 }, { 0x1036, 0x1037 }, { 0x1039, 0x1039 },
 160     { 0x1058, 0x1059 }, { 0x1160, 0x11FF }, { 0x135F, 0x135F },
 161     { 0x1712, 0x1714 }, { 0x1732, 0x1734 }, { 0x1752, 0x1753 },
 162     { 0x1772, 0x1773 }, { 0x17B4, 0x17B5 }, { 0x17B7, 0x17BD },
 163     { 0x17C6, 0x17C6 }, { 0x17C9, 0x17D3 }, { 0x17DD, 0x17DD },
 164     { 0x180B, 0x180D }, { 0x18A9, 0x18A9 }, { 0x1920, 0x1922 },
 165     { 0x1927, 0x1928 }, { 0x1932, 0x1932 }, { 0x1939, 0x193B },
 166     { 0x1A17, 0x1A18 }, { 0x1B00, 0x1B03 }, { 0x1B34, 0x1B34 },
 167     { 0x1B36, 0x1B3A }, { 0x1B3C, 0x1B3C }, { 0x1B42, 0x1B42 },
 168     { 0x1B6B, 0x1B73 }, { 0x1DC0, 0x1DCA }, { 0x1DFE, 0x1DFF },
 169     { 0x200B, 0x200F }, { 0x202A, 0x202E }, { 0x2060, 0x2063 },
 170     { 0x206A, 0x206F }, { 0x20D0, 0x20EF }, { 0x302A, 0x302F },
 171     { 0x3099, 0x309A }, { 0xA806, 0xA806 }, { 0xA80B, 0xA80B },
 172     { 0xA825, 0xA826 }, { 0xFB1E, 0xFB1E }, { 0xFE00, 0xFE0F },
 173     { 0xFE20, 0xFE23 }, { 0xFEFF, 0xFEFF }, { 0xFFF9, 0xFFFB },
 174     { 0x10A01, 0x10A03 }, { 0x10A05, 0x10A06 }, { 0x10A0C, 0x10A0F },
 175     { 0x10A38, 0x10A3A }, { 0x10A3F, 0x10A3F }, { 0x1D167, 0x1D169 },
 176     { 0x1D173, 0x1D182 }, { 0x1D185, 0x1D18B }, { 0x1D1AA, 0x1D1AD },
 177     { 0x1D242, 0x1D244 }, { 0xE0001, 0xE0001 }, { 0xE0020, 0xE007F },
 178     { 0xE0100, 0xE01EF }
 179   };
 180
 181   /* test for 8-bit control characters */
 182   if (ucs == 0)
 183     return 0;
 184   if (ucs < 32 || (ucs >= 0x7f && ucs < 0xa0))
 185     return -1;
 186
 187   /* binary search in table of non-spacing characters */
 188   if (bisearch(ucs, combining,
 189                sizeof(combining) / sizeof(struct interval) - 1))
 190     return 0;
 191
 192   /* if we arrive here, ucs is not a combining or C0/C1 control character */
 193
 194   return 1 +
 195     (ucs >= 0x1100 &&
 196      (ucs <= 0x115f ||                    /* Hangul Jamo init. consonants */
 197       ucs == 0x2329 || ucs == 0x232a ||
 198       (ucs >= 0x2e80 && ucs <= 0xa4cf &&
 199        ucs != 0x303f) ||                  /* CJK ... Yi */
 200       (ucs >= 0xac00 && ucs <= 0xd7a3) || /* Hangul Syllables */
 201       (ucs >= 0xf900 && ucs <= 0xfaff) || /* CJK Compatibility Ideographs */
 202       (ucs >= 0xfe10 && ucs <= 0xfe19) || /* Vertical forms */
 203       (ucs >= 0xfe30 && ucs <= 0xfe6f) || /* CJK Compatibility Forms */
 204       (ucs >= 0xff00 && ucs <= 0xff60) || /* Fullwidth Forms */
 205       (ucs >= 0xffe0 && ucs <= 0xffe6) ||
 206       (ucs >= 0x20000 && ucs <= 0x2fffd) ||
 207       (ucs >= 0x30000 && ucs <= 0x3fffd)));
 208 }
 209
 210
 211 int mk_wcswidth(const wchar_t *pwcs, size_t n)
 212 {
 213   int w, width = 0;
 214
 215   for (;*pwcs && n-- > 0; pwcs++)
 216     if ((w = mk_wcwidth(*pwcs)) < 0)
 217       return -1;
 218     else
 219       width += w;
 220
 221   return width;
 222 }
 223
 224
 225 /*
 226  * The following functions are the same as mk_wcwidth() and
 227  * mk_wcswidth(), except that spacing characters in the East Asian
 228  * Ambiguous (A) category as defined in Unicode Technical Report #11
 229  * have a column width of 2. This variant might be useful for users of
 230  * CJK legacy encodings who want to migrate to UCS without changing
 231  * the traditional terminal character-width behaviour. It is not
 232  * otherwise recommended for general use.
 233  */
 234 int mk_wcwidth_cjk(wchar_t ucs)
 235 {
 236   /* sorted list of non-overlapping intervals of East Asian Ambiguous
 237    * characters, generated by "uniset +WIDTH-A -cat=Me -cat=Mn -cat=Cf c" */
 238   static const struct interval ambiguous[] = {
 239     { 0x00A1, 0x00A1 }, { 0x00A4, 0x00A4 }, { 0x00A7, 0x00A8 },
 240     { 0x00AA, 0x00AA }, { 0x00AE, 0x00AE }, { 0x00B0, 0x00B4 },
 241     { 0x00B6, 0x00BA }, { 0x00BC, 0x00BF }, { 0x00C6, 0x00C6 },
 242     { 0x00D0, 0x00D0 }, { 0x00D7, 0x00D8 }, { 0x00DE, 0x00E1 },
 243     { 0x00E6, 0x00E6 }, { 0x00E8, 0x00EA }, { 0x00EC, 0x00ED },
 244     { 0x00F0, 0x00F0 }, { 0x00F2, 0x00F3 }, { 0x00F7, 0x00FA },
 245     { 0x00FC, 0x00FC }, { 0x00FE, 0x00FE }, { 0x0101, 0x0101 },
 246     { 0x0111, 0x0111 }, { 0x0113, 0x0113 }, { 0x011B, 0x011B },
 247     { 0x0126, 0x0127 }, { 0x012B, 0x012B }, { 0x0131, 0x0133 },
 248     { 0x0138, 0x0138 }, { 0x013F, 0x0142 }, { 0x0144, 0x0144 },
 249     { 0x0148, 0x014B }, { 0x014D, 0x014D }, { 0x0152, 0x0153 },
 250     { 0x0166, 0x0167 }, { 0x016B, 0x016B }, { 0x01CE, 0x01CE },
 251     { 0x01D0, 0x01D0 }, { 0x01D2, 0x01D2 }, { 0x01D4, 0x01D4 },
 252     { 0x01D6, 0x01D6 }, { 0x01D8, 0x01D8 }, { 0x01DA, 0x01DA },
 253     { 0x01DC, 0x01DC }, { 0x0251, 0x0251 }, { 0x0261, 0x0261 },
 254     { 0x02C4, 0x02C4 }, { 0x02C7, 0x02C7 }, { 0x02C9, 0x02CB },
 255     { 0x02CD, 0x02CD }, { 0x02D0, 0x02D0 }, { 0x02D8, 0x02DB },
 256     { 0x02DD, 0x02DD }, { 0x02DF, 0x02DF }, { 0x0391, 0x03A1 },
 257     { 0x03A3, 0x03A9 }, { 0x03B1, 0x03C1 }, { 0x03C3, 0x03C9 },
 258     { 0x0401, 0x0401 }, { 0x0410, 0x044F }, { 0x0451, 0x0451 },
 259     { 0x2010, 0x2010 }, { 0x2013, 0x2016 }, { 0x2018, 0x2019 },
 260     { 0x201C, 0x201D }, { 0x2020, 0x2022 }, { 0x2024, 0x2027 },
 261     { 0x2030, 0x2030 }, { 0x2032, 0x2033 }, { 0x2035, 0x2035 },
 262     { 0x203B, 0x203B }, { 0x203E, 0x203E }, { 0x2074, 0x2074 },
 263     { 0x207F, 0x207F }, { 0x2081, 0x2084 }, { 0x20AC, 0x20AC },
 264     { 0x2103, 0x2103 }, { 0x2105, 0x2105 }, { 0x2109, 0x2109 },
 265     { 0x2113, 0x2113 }, { 0x2116, 0x2116 }, { 0x2121, 0x2122 },
 266     { 0x2126, 0x2126 }, { 0x212B, 0x212B }, { 0x2153, 0x2154 },
 267     { 0x215B, 0x215E }, { 0x2160, 0x216B }, { 0x2170, 0x2179 },
 268     { 0x2190, 0x2199 }, { 0x21B8, 0x21B9 }, { 0x21D2, 0x21D2 },
 269     { 0x21D4, 0x21D4 }, { 0x21E7, 0x21E7 }, { 0x2200, 0x2200 },
 270     { 0x2202, 0x2203 }, { 0x2207, 0x2208 }, { 0x220B, 0x220B },
 271     { 0x220F, 0x220F }, { 0x2211, 0x2211 }, { 0x2215, 0x2215 },
 272     { 0x221A, 0x221A }, { 0x221D, 0x2220 }, { 0x2223, 0x2223 },
 273     { 0x2225, 0x2225 }, { 0x2227, 0x222C }, { 0x222E, 0x222E },
 274     { 0x2234, 0x2237 }, { 0x223C, 0x223D }, { 0x2248, 0x2248 },
 275     { 0x224C, 0x224C }, { 0x2252, 0x2252 }, { 0x2260, 0x2261 },
 276     { 0x2264, 0x2267 }, { 0x226A, 0x226B }, { 0x226E, 0x226F },
 277     { 0x2282, 0x2283 }, { 0x2286, 0x2287 }, { 0x2295, 0x2295 },
 278     { 0x2299, 0x2299 }, { 0x22A5, 0x22A5 }, { 0x22BF, 0x22BF },
 279     { 0x2312, 0x2312 }, { 0x2460, 0x24E9 }, { 0x24EB, 0x254B },
 280     { 0x2550, 0x2573 }, { 0x2580, 0x258F }, { 0x2592, 0x2595 },
 281     { 0x25A0, 0x25A1 }, { 0x25A3, 0x25A9 }, { 0x25B2, 0x25B3 },
 282     { 0x25B6, 0x25B7 }, { 0x25BC, 0x25BD }, { 0x25C0, 0x25C1 },
 283     { 0x25C6, 0x25C8 }, { 0x25CB, 0x25CB }, { 0x25CE, 0x25D1 },
 284     { 0x25E2, 0x25E5 }, { 0x25EF, 0x25EF }, { 0x2605, 0x2606 },
 285     { 0x2609, 0x2609 }, { 0x260E, 0x260F }, { 0x2614, 0x2615 },
 286     { 0x261C, 0x261C }, { 0x261E, 0x261E }, { 0x2640, 0x2640 },
 287     { 0x2642, 0x2642 }, { 0x2660, 0x2661 }, { 0x2663, 0x2665 },
 288     { 0x2667, 0x266A }, { 0x266C, 0x266D }, { 0x266F, 0x266F },
 289     { 0x273D, 0x273D }, { 0x2776, 0x277F }, { 0xE000, 0xF8FF },
 290     { 0xFFFD, 0xFFFD }, { 0xF0000, 0xFFFFD }, { 0x100000, 0x10FFFD }
 291   };
 292
 293   /* binary search in table of non-spacing characters */
 294   if (bisearch(ucs, ambiguous,
 295                sizeof(ambiguous) / sizeof(struct interval) - 1))
 296     return 2;
 297
 298   return mk_wcwidth(ucs);
 299 }
 300
 301
 302 int mk_wcswidth_cjk(const wchar_t *pwcs, size_t n)
 303 {
 304   int w, width = 0;
 305
 306   for (;*pwcs && n-- > 0; pwcs++)
 307     if ((w = mk_wcwidth_cjk(*pwcs)) < 0)
 308       return -1;
 309     else
 310       width += w;
 311
 312   return width;
 313 }